Create app.py

app.py

@@ -0,0 +1,255 @@
+import gradio as gr
+import numpy as np
+import pickle
+import os
+import sqlite3
+
+
+def create_database():
+    # Connecting to the SQLite database
+    conn = sqlite3.connect('churndatabase.db')
+    cursor = conn.cursor()
+
+    # Create customer_churn table
+    cursor.execute("""
+    CREATE TABLE customer_churn (
+        customer_id INT PRIMARY KEY,
+        gender VARCHAR(10),
+        age INT,
+        marital_status VARCHAR(10),
+        dependents INT,
+        contract_type VARCHAR(10),
+        internet_service VARCHAR(20),
+        phone_service VARCHAR(3),
+        multiple_lines VARCHAR(3),
+        online_security VARCHAR(3),
+        online_backup VARCHAR(3),
+        device_protection VARCHAR(3),
+        tech_support VARCHAR(3),
+        streaming_tv VARCHAR(3),
+        streaming_movies VARCHAR(3),
+        monthly_charges NUMBER(8, 2),
+        total_charges NUMBER(10, 2),
+        churn_status VARCHAR(3)
+    )
+    """)
+
+    # Create customer_churn_full table
+    cursor.execute("""
+    CREATE TABLE customer_churn_full (
+        customer_id INT PRIMARY KEY,
+        gender VARCHAR(10),
+        age INT,
+        marital_status VARCHAR(10),
+        dependents INT,
+        contract_type VARCHAR(10),
+        internet_service VARCHAR(20),
+        phone_service VARCHAR(3),
+        multiple_lines VARCHAR(3),
+        online_security VARCHAR(3),
+        online_backup VARCHAR(3),
+        device_protection VARCHAR(3),
+        tech_support VARCHAR(3),
+        streaming_tv VARCHAR(3),
+        streaming_movies VARCHAR(3),
+        monthly_charges NUMBER(8, 2),
+        total_charges NUMBER(10, 2),
+        churn_status VARCHAR(3),
+        call_duration_minutes INT,
+        latitude NUMBER(9, 6),
+        longitude NUMBER(9, 6)
+    )
+    """)
+
+    # Insert data into customer_churn table
+    cursor.execute("""
+    INSERT INTO customer_churn
+    SELECT
+        TRUNC(RANDOM() * 1000000) AS customer_id,
+        CASE WHEN RANDOM() < 0.5 THEN 'Male' ELSE 'Female' END AS gender,
+        TRUNC(RANDOM() * 60 + 18) AS age,
+        CASE WHEN RANDOM() < 0.5 THEN 'Married' ELSE 'Single' END AS marital_status,
+        TRUNC(RANDOM() * 5) AS dependents,
+        CASE WHEN RANDOM() < 0.5 THEN 'Monthly' ELSE 'Yearly' END AS contract_type,
+        CASE WHEN RANDOM() < 0.5 THEN 'DSL' ELSE 'Fiber Optic' END AS internet_service,
+        CASE WHEN RANDOM() < 0.5 THEN 'Yes' ELSE 'No' END AS phone_service,
+        CASE WHEN RANDOM() < 0.5 THEN 'Yes' ELSE 'No' END AS multiple_lines,
+        CASE WHEN RANDOM() < 0.5 THEN 'Yes' ELSE 'No' END AS online_security,
+        CASE WHEN RANDOM() < 0.5 THEN 'Yes' ELSE 'No' END AS online_backup,
+        CASE WHEN RANDOM() < 0.5 THEN 'Yes' ELSE 'No' END AS device_protection,
+        CASE WHEN RANDOM() < 0.5 THEN 'Yes' ELSE 'No' END AS tech_support,
+        CASE WHEN RANDOM() < 0.5 THEN 'Yes' ELSE 'No' END AS streaming_tv,
+        CASE WHEN RANDOM() < 0.5 THEN 'Yes' ELSE 'No' END AS streaming_movies,
+        ROUND(RANDOM() * 100, 2) AS monthly_charges,
+        ROUND(RANDOM() * 1000, 2) AS total_charges,
+        CASE WHEN RANDOM() < 0.2 THEN 'Yes' ELSE 'No' END AS churn_status
+    FROM
+        (SELECT 1)
+    CROSS JOIN
+        (SELECT 1)
+    LIMIT
+        500
+    """)
+
+    # Insert data into customer_churn_full table
+    cursor.execute("""
+    INSERT INTO customer_churn_full
+    SELECT
+        TRUNC(RANDOM() * 1000000) AS customer_id,
+        CASE WHEN RANDOM() < 0.5 THEN 'Male' ELSE 'Female' END AS gender,
+        TRUNC(RANDOM() * 60 + 18) AS age,
+        CASE WHEN RANDOM() < 0.5 THEN 'Married' ELSE 'Single' END AS marital_status,
+        TRUNC(RANDOM() * 5) AS dependents,
+        CASE WHEN RANDOM() < 0.5 THEN 'Monthly' ELSE 'Yearly' END AS contract_type,
+        CASE WHEN RANDOM() < 0.5 THEN 'DSL' ELSE 'Fiber Optic' END AS internet_service,
+        CASE WHEN RANDOM() < 0.5 THEN 'Yes' ELSE 'No' END AS phone_service,
+        CASE WHEN RANDOM() < 0.5 THEN 'Yes' ELSE 'No' END AS multiple_lines,
+        CASE WHEN RANDOM() < 0.5 THEN 'Yes' ELSE 'No' END AS online_security,
+        CASE WHEN RANDOM() < 0.5 THEN 'Yes' ELSE 'No' END AS online_backup,
+        CASE WHEN RANDOM() < 0.5 THEN 'Yes' ELSE 'No' END AS device_protection,
+        CASE WHEN RANDOM() < 0.5 THEN 'Yes' ELSE 'No' END AS tech_support,
+        CASE WHEN RANDOM() < 0.5 THEN 'Yes' ELSE 'No' END AS streaming_tv,
+        CASE WHEN RANDOM() < 0.5 THEN 'Yes' ELSE 'No' END AS streaming_movies,
+        ROUND(RANDOM() * 100, 2) AS monthly_charges,
+        ROUND(RANDOM() * 1000, 2) AS total_charges,
+        CASE WHEN RANDOM() < 0.2 THEN 'Yes' ELSE 'No' END AS churn_status,
+        TRUNC(RANDOM() * 1200) AS call_duration_minutes,
+        ROUND(RANDOM() * 180 - 90, 6) AS latitude,
+        ROUND(RANDOM() * 360 - 180, 6) AS longitude
+    FROM
+        (SELECT 1)
+    CROSS JOIN
+        (SELECT 1)
+    LIMIT
+        500
+    """)
+
+    conn.commit()
+    conn.close()
+
+create_database()
+
+#Loading ML Components
+DIRPATH = os.path.dirname(os.path.realpath(__file__))
+ml_join= os.path.join(DIRPATH,'ml_components.pkl')
+with open(ml_join,"rb") as f:
+    loaded_model_components = pickle.load(f)
+
+# Extracting ML components
+model = loaded_model_components['model']
+encoder = loaded_model_components['encoder']
+scaler = loaded_model_components['scaler']
+
+#To predict churn
+def predict_churn(gender, SeniorCitizen, Partner, Dependents, PhoneService, MultipleLines,
+                InternetService, OnlineSecurity, OnlineBackup, DeviceProtection, TechSupport,
+                StreamingTV, StreamingMovies, Contract, PaperlessBilling, PaymentMethod,
+                tenure, MonthlyCharges, TotalCharges):
+    
+    #Encoding categorical features
+    encoded_data = encoder.transform([[gender, SeniorCitizen, Partner, Dependents, PhoneService, MultipleLines,
+                                    InternetService, OnlineSecurity, OnlineBackup, DeviceProtection, TechSupport,
+                                    StreamingTV, StreamingMovies, Contract, PaperlessBilling, PaymentMethod]])
+    
+    num_data = np.array([[tenure, MonthlyCharges, TotalCharges]])
+    scaled_num_data = scaler.transform(num_data)
+    
+    combined_data = np.hstack((encoded_data, scaled_num_data))
+        
+    #Making predictions using fitted model
+    model_output = model.predict_proba(combined_data)
+    prob_churn = float(model_output[0][1])
+    return {'Churn Probability': prob_churn}
+
+with gr.Blocks(css="""
+        background-color: #6db05d;
+        padding: 2rem;
+        border: 2px solid #e86347;
+    }
+    .gradio-input {
+        border: 1px solid #ccc000;
+        padding: 0.5rem;
+        border-radius: 0.5rem;
+        font-size: 1rem;
+    }
+    .gradio-label {
+        font-weight: bold;
+        margin-bottom: 0.5rem;
+    }
+    .gradio-markdown h1 {
+        font-size: 4rem;
+        font-family: 'Georgia', serif;
+        margin-top: 0;
+        margin-bottom: 0.5rem;
+        color: white;
+        background-color: #1E90FF;
+        padding: 1rem;
+        border-radius: 0.5rem;
+        text-align: center;
+    }
+    .gradio-markdown h2 {
+        font-size: 2rem;
+        font-family: 'DejaVu Sans Mono', monospace;
+        text-align: center;
+        margin-top: 0;
+        margin-bottom: 0.5rem;
+        color: white;
+        background-color: #4169E1;
+        padding: 0.5rem 1rem;
+        border-radius: 0.5rem;
+    }
+    .gradio-accordion {
+        background-color: #6db05d;
+        border-radius: 0.75rem;
+        padding: 1rem;
+        margin-bottom: 1rem;
+        text-align: left;
+    }
+""") as test:
+
+    gr.Markdown("# CHURN PREDICTION")
+    gr.Markdown("## Using machine learning and Oracle SQL")
+    gr.Markdown("### By Hitesh Beeraka and Yash Singhvi")
+    with gr.Row():
+        with gr.Column(scale=2):
+            with gr.Accordion(label="Personal Information", open=False):
+                gender = gr.Radio(choices=['Male', 'Female'], label='Gender', interactive=True)
+                SeniorCitizen = gr.Radio(choices=['Yes', 'No'], label='SeniorCitizen', interactive=True)
+                Partner = gr.Radio(choices=['Yes', 'No'], label='Partner', interactive=True)
+                Dependents = gr.Radio(choices=['Yes', 'No'], label='Dependents', interactive=True)
+        with gr.Column(scale=2):
+            with gr.Accordion(label="Phone and Internet Services", open=False):
+                PhoneService = gr.Radio(choices=['Yes', 'No'], label='PhoneService', interactive=True)
+                MultipleLines = gr.Radio(choices=['Yes', 'No'], label='MultipleLines', interactive=True)
+                InternetService = gr.Radio(choices=['Fiber optic', 'No', 'DSL'], label='InternetService', interactive=True)
+    with gr.Row():
+        with gr.Column(scale=2):
+            with gr.Accordion(label="Security and Support Services", open=False):
+                OnlineSecurity = gr.Radio(choices=['Yes', 'No'], label='OnlineSecurity', interactive=True)
+                OnlineBackup = gr.Radio(choices=['Yes', 'No'], label='OnlineBackup', interactive=True)
+                DeviceProtection = gr.Radio(choices=['Yes', 'No'], label='DeviceProtection', interactive=True)
+                TechSupport = gr.Radio(choices=['Yes', 'No'], label='TechSupport', interactive=True)
+        with gr.Column(scale=2):
+            with gr.Accordion(label="Entertainment Services", open=False):
+                StreamingTV = gr.Radio(choices=['Yes', 'No'], label='StreamingTV', interactive=True)
+                StreamingMovies = gr.Radio(choices=['Yes', 'No'], label='StreamingMovies', interactive=True)
+    with gr.Row():
+        with gr.Column(scale=2):
+            with gr.Accordion(label="Contract and Billing Information", open=False):
+                Contract = gr.Radio(choices=['Month-to-month', 'One year', 'Two year'], label='Contract', interactive=True)
+                PaperlessBilling = gr.Radio(choices=['Yes', 'No'], label='PaperlessBilling', interactive=True)
+                PaymentMethod = gr.Radio(choices=['Electronic check', 'Mailed check', 'Credit card (automatic)', 'Bank transfer (automatic)'], label='PaymentMethod', interactive=True)
+        with gr.Column(scale=2):
+            with gr.Accordion(label="Charges and Tenure", open=False):
+                tenure = gr.Number(label='Tenure', interactive=True)
+                MonthlyCharges = gr.Number(label='MonthlyCharges', interactive=True)
+                TotalCharges = gr.Number(label='TotalCharges', interactive=True)
+    
+    output = gr.Label(label="Churn Probability")
+    submit_btn = gr.Button("Predict")
+    submit_btn.click(
+        fn=predict_churn,
+        inputs=[gender, SeniorCitizen, Partner, Dependents, PhoneService, MultipleLines, InternetService, OnlineSecurity, OnlineBackup, DeviceProtection, TechSupport, StreamingTV, StreamingMovies, Contract, PaperlessBilling, PaymentMethod, tenure, MonthlyCharges, TotalCharges],
+        outputs=output
+    )
+test.launch(inbrowser=True, show_error=True,share=True)